Qwen-7B-Chat
by Qwen
7.7B params · text-generation · 788 likes · 172.7k downloads
Qwen-7B-Chat is a 7.7B parameter model. At Q4 quantization it requires 4GB of VRAM. It runs comfortably on GeForce RTX 4090 (170 tok/s), GeForce RTX 5090 (255 tok/s), M4 Max 128GB (62 tok/s).
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 170 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 255 tok/s | comfortable |
| M4 Max 128GB | 128GB | 62 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 31 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 31 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 311 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 626 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 150 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 124 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 120 tok/s | comfortable |